Factors Affecting Power - Effect size, Variability, Sample Size (Module 1 8 7)

[misslollipoppa]

Thanks so much! This video saved me one hour before my statistics test!

[tomp]

Thank you for a great explanation on a confusing topic.
Is it possible to have "too much" power?
For example, when doing a Chi Square test for Benford's Law, with N=3140 and p-value was calculated to be .015. The critique was there was "too much" power. That the test was too sensitive to minor variations of several leading digits because of the high sample size.

[jangofett]

Is the variability you were talking about the same thing as significance level? In my book it says increasing the significance level will increase the power of the test, and that it has something to do with the probability of a T2 error, but I'm quite confused as to how everything is related. Any help would be greatly appreciated. Thanks for making this video!

[giacmon]

I don't fully understand how effect size is considered a factor for improving power of a test. I get it that mathematically the effect is at the numerator of the formula and therefore the larger the effect size the larger the Z statistic on average will be, and therefore the more power. However I do not understand how you, as a scientist, can influence this. It seems to me that effect size is a result of an experiment, not a parameter. Example: my null hypothesis is that the mean is a certain value X0. I run a test and I get a value Xa for my alternative hypothesis. And that's it. I have the effect size. I don't get to choose it, do I? I hope I made myself clear. Thanks for this great video.

[daniellechen]

Larger effect size, smaller variance, higher power, lower sample size
Larger effect size, smaller variance, larger sample size, higher power